The UK's Autonomous Vehicle Permit Process Explained

Welcome to Autonomy Insiders. The show where global industry leaders in autonomous driving unpack their real world insights. To have a supervised driverless vehicle on the road, you can just do that. There is no state-to-to-rate requirement behind that. They are taxi drivers, Uber drivers, over at Wave. There's no legal difference between the command service available on the Uber app to take on a supervised autonomous vehicle ride and a standard Uber vehicle ride. Politically they can't stop at this forever, but actually legally they can actually delay this for a very long time, ultimately for London forever, because they are the contender. I'm your host Daniel and on September 3rd, London got its first robot taxi ride heading service. Open on the Uber app and you might get a Ford Mustang match E driven by Wave's AI, but there's still a license driver behind the wheel end. Knowing that driver is a completely different legal process as of now, has no company had received that permit. So how does it actually work, who approves what and why is it taking longer than many expect? My guest today is Tim Sirichick. Tim co-founded the Institute for Driverless Transport and Independent UK Research Body that studies what autonomous vehicles mean for Britain, and he also rides self-driving insights and his latest piece breaks down exactly how you get permission to run a robot taxi service in London. So Tim, welcome to the show. Great to be here. Thanks for the invite. Tim, you started self-driving insights and then co-founded IFTT. So what gap in the UK conversation made you think a news letter was not enough? Yeah, so great question. The UK has had incredible research and interest in autonomous vehicles over the past 20 years you could say even longer, and there's great work being done right now in the government. Right now we're processing through the legislation as to how we can admit any form of autonomous vehicle service in the UK, and there's also great academic work. But outside of academic work and the government, there was not enough industry working our view. There were some academic and industry partnerships, but not a lot of considerations as to how to make this a technology integrated in the economy, which areas of our economy, which cities would benefit the most from autonomous vehicles, and what would be the long-term implications of that. There was quite a lot of conversations on business cases and safety cases in the mobility sector, but we wanted to think, well, how does the world look different 20 years from now? A sort of analogy that we like to describe is it would have been great if we had an institute for AI in 2015. If we could actually think, oh, what's going to happen in 2030, 2026 about AI, and then maybe some of us would have been able to do research about labor transitions or adoption in different economic sectors. So that was the motivation behind IFDT. We have the launch event in April, attended by about 200, and ever since then we've been more and more busy with questions and answers and research, but we've been very happy to take on here. Sounds good. And who is IFDT then really forg is it the government, is it the industry, or the general public? I would say the public, right? Ultimately, we want to benefit the UK and benefit economic outcomes, social outcomes, also make sure that the geopolitical conversation is made more salient, and also we are really interested in the job displacement question. Those are our four areas, all of them are areas that benefit the government, they benefit the general public, they also benefit the industry to successfully roll out in a way that minimises disruption and rings out economic benefits. And we're not focused entirely in the UK either, right? A lot of our interest has also been the public and transport operators from outside the UK. So we're open to that conversation, we're open to producing research that hopefully is a general purpose good. And just as I wrote about the permitting licenses here in the UK, I might in the future write about permitting licenses in other countries if I think there will be benefits to do so. What motivated me to write this particular piece was because I personally thought there wasn't a very good explainer of the government legislation, there's a lot of gov.uk writing, which is an excellent website and all the details that you need will be there. But for something that like an everyday person or someone really interested in the industry, but not very familiar with all the legal jargon or the legal agencies involved, this was a great way to move forward. And one question I want to preempt as well is who are we and who funds us. This is a question that we get quite often. We are an independent research organisation. We're not part of any academic organisation and we currently feel anthropically funded. So we've previously received a grant form of a Mercator Centre known as emergent ventures. This is a one off grant, no strings attached, just to help us get off the ground. And now we're planning to write commission research but also apply for other grant processes that hopefully will allow this small organisation to continue in the near future. So yeah, hopefully that addresses that. And you mentioned your piece about the permitting process already. So most if we cover it just about the technology. So why should someone who builds a software care about a UK permit process just as much as about the technology? Well, if you care about the technology being used, if you care about outcomes, then the permitting technology and the licensing agreements are just as important as the software. We can have the software, we can have a great operation going in test cases, in private land and private roads, but we want passengers on the road and we need more than just software to make that work. We need a lot of safety assurances. We need a lot of understanding of who is responsible in any given moment. We need integration with other forms of transport and other forms of action that happens on the road. So for instance, we need to know what to do when there is a emergency vehicle driving to pass the software, not just from the software side or driving past the robot taxi for instance, not just on the software side, but also from a operational side. So those are the motivations. Not every software engineer has to be interested, but if you're interested in the outcomes, this is my recommendation. All right. So let's clear up the basics first. So everyone listening has also the same vocabulary. So in the UK law, what is the difference between a super rice autonomous ride and a driverless one? Right. So a supervised autonomous ride is a ride where the user is still in control. That means that there is still a driver being responsible for the behavior of the vehicle. This was defined in the automated vehicles act. And even if the vehicle is driving by itself all of the time, because it is legally defined as a supervised ride, the person and the safety driver will be responsible. A driver's vehicle doesn't have that. Currently, there are no driverless vehicles legally in the UK operating as driverless vehicles. They have to be placed as supervised vehicles, because they don't have a permit to operate as driverless vehicles. So that's the distinction and the conversation will have about the permitting scheme will define how we can switch from a supervised into a driverless vehicle. Yeah. This makes sense. And this legal difference comes also from two different laws, which probably confuses a lot of people. So there are two key laws. It's the automated electric vehicles act from 2018 and the automated vehicle act from 2024. So what does each one do and why does the 2018 act still matter in 2026? Yeah. Great question. Automated and electric vehicles act of 2018 provided the first sort of insurance and definitions of what an autonomous vehicle is and when would the responsibility be set? It would ever be a payout. This is actually really important even for any trials or tests so that even if there are no public uses of autonomous vehicles, we can conduct trials and if something goes wrong, all the money gets passed around legally as clearly defined in practice. I think in the vast majority of cases, we've been always in a user and control environment. But it also set up a very important foundation for the automated and vehicles act in 2024 and for these trials that we see today. The automated vehicles act of 2024 wants to define autonomous vehicles permanently in the UK for all use cases. So that all explain who will be responsible for licensing, who will be responsible for operating. It will define any sort of incident investigation. Everything that you can consider the state of a government being responsible or interested in when there's an autonomous vehicle on the road, this is where it's going to fight. the law has passed. in 2024 but the way British law has worked for a while and the way that even European Union law works is that even when the act gets passed there's a lot of definitions and a lot of secondary legislation or clarifications that have to be published and that happens often in conjunction with industry and right now this regulatory environment will be completed or scheduled to be completed by 2027 so by the end of next year we have everything legally defined in autonomous vehicles right now when this middle stage and the laws are the one thing but the people who apply them are another so can you walk us through the cast so the Department of Transport, the VCA, the DVSA and TFL so who does what? Yes and there's quite a lot of acronyms here as always you can reference gov.uk or search visa and DVSA transport over like so let's start from the national government the national government sits in London Westminster and has several departments the one we're interested in today is the Department for Transport the person responsible for the Department for Transport is a politician will currently held by Heidi Alexander the role of Secretary of State so whenever you see SOS it actually doesn't mean that anyone's in trouble in transport it means that we're referring to the Secretary of State the most important politician who officially will be the person to grant the license in practice the Department for Transport has different agencies has different organizations but help do the decision making there's a driver vehicle and standards authority which does the vast majority of the permitting process they will accept the application from an operator to operate a private hire like vehicle service or a robot taxi service and they will process it they will request a safety case and they will be the ones to give approval on behalf of the Secretary of State at times another agency will get involved this is the vehicle certifications agency VCA both the DVSA and the VCA are part of the Department for Transport so ultimately they'll be working very closely together they'll be in fact working parallel to each other in different applications and those are the major players from the government side and the central government then there is the local transport operators and the local consenting authorities the organizations that have the local voice in mind whenever a permit gets licensed in the case of London that will be transport the London the operator that operates the London underground the tube network as it's known and also the bus network and light rail services in other places in the UK it might be other transport operators or transport authorities to be precise transport authorities don't always include operators like TFL but instead could also include local government so the transport authority in the city of leads will be the city of leads council so as these licenses get applied in other places in the UK there might be different local authorities or local voices or local consenting authorities that will be in play and the government also has fast-tracked the pilot scheme from late 2027 to spring 2026 and what did fast tracking actually change and what did it maybe also leave the same fast-tracking I'm allowed for the automated passenger service permitting scheme to be opened the automated passenger service permitting scheme is the legal pathway through which any company that wants to operate an autonomous vehicle in a bus like service or a taxi like service can get the legal permission to do so including fully driverless services so you'll be legally allowed to take out a driver have a remote operator or not even have a remote operator if you can make the safety case for it and or supervisor and have these on public roads currently this is the only legal pathway to do so it is a introduction of a part of the regulatory environment earlier so that full regulatory considerations for instance nationwide permitting haven't been considered yet that we've only started considering these local services for very particular use cases so for instance there isn't a legal pathway for semi public or private local or government owned private autonomous vehicles for instance so companies can apply for the automated passenger service but it is less clear as to for instance if you don't want to do a bus like service and you want to use this for commercial operations with some humans involved then there might be other pathways to consider and what happens to such an pilot permit when the permanent framework arrives the current framework of automated passenger service permits are by nature temporary so they have to be renewed and when the when the full regulatory environment arrives they'll keep that same structure essentially they'll be maybe more clarification as to how quickly you can get a license but the same process will be at play so the same actors the same paperwork will be important let's follow a company through the process in order so it starts long before any passenger gets in so way more or wave and also a polygon they're all testing in London if there's no trial permit so how does that work so to have a supervised driverless vehicle on the road you can just do that there is a no state-to-total requirements behind that if I personally buy myself a Tesla and I have a driver assist features whatever they may be the lower features of a full self-driving supervised which isn't yet legal in the UK I can do that because I am ultimately responsible as the driver for this so in these trials the companies are collecting data across London with however many vehicles they want with however many supervisors or safety drivers and ultimately there's only a guidance and a recommendation by the government published to support this to make these trials as conducive with government preferences as possible known as the collaborative practice as mentioned it's not statutory but it allows these companies to gather evidence and conduct these rides in a way that's legally safe and also might help with any future applications for autonomous passenger services when they want to take out the driver eventually and you mentioned a code of practices non-stitutory but what happens if a company then ignores it well most ways that they can ignore it would involve criminal activity it has been long defined for instance passing a red light so the standard practices of finding would occur points on a license eventually there might be considered it would fare badly as well on any applications for autonomous passenger service permits but ultimately we haven't yet got into that situation later on as we talk about the permitting scheme we notice there are elements of judgment there's almost elements of trust both from the DBSA and from TFL towards the operators so here the relationship would get damaged even though there might not be a legal mechanism to directly punish the company to the scale that maybe some people would consider important if this code would be violated yeah and I think one part of the code is also a safety case that is expected and you mentioned several times already so in plain terms what goes into that or into a good safety case and also who actually reads that safety case that so the safety case is a form of evidence which explains why a particular autonomous vehicle is allowed to operate on the road or should be allowed to operate on the roads and that it's safe to do so it includes not just the definition of the software and what can the software do in which environment but also explanations as to what is the company doing to ensure safety from an operational standpoint cyber security considerations vetting and screening of the safety drivers considerations of interactions with world authorities and all of this can be updated changed at any point because the operators should provide the most recent evidence and these trials can help in updating the safety case any UK authority in this case the DBSA and also the VCA would both be readers of the safety safety cases the VCA the vehicle certification agency would be interested most in the software and in the vehicle design itself so let's say there is a vehicle designed that submitted for a permit that doesn't have a steering wheel then there would have to be specific considerations of okay how can we be sure that remote operations are as reliable as we want them to be we assume reliability of control by a human with a steering wheel but there will be additional evidence required for this particular case on control of the vehicle the DBSA will be more interested in the operational side but both would want evidence and evidence from the UK in particular will be excellent for these organizations because they want to know safety on British roads safety around edge cases on roundabouts, but might be other evidence to gather in the United States, for instance. Make sense. And now you mentioned several times the VCA and one thing that VCA does, also the vehicle special order. So can you explain what is the vehicle special order and also when does the company need one? Yeah. The vehicle special order is a license that make sure that the vehicle is safe. Obviously, in the case of autonomous vehicles, it will consider things like do we actually need rearview mirrors for safety here in this case? So the VSO will be a new application for any new design of a vehicle. So in my view, the Ford Mustangs probably did not require a VSO, but I'm not entirely sure, but if we were to bring out a ZUX vehicle on the British roads, a VSO would be required before you're even allowed to do the trials. This is different to the code described earlier. This is a legal requirement that has always been the case. And do you also have an estimate how long such a VSO usually takes? From what I understand, this is a matter of weeks for set blueprints. I think the more complex the design, this could be a matter of months, but at first, it could be less than a month, especially when it's a small notification. And looking at the company, so how does a company know it has tested enough in a city to apply for the driverless permit? Is is there a specific threshold in the laws or does the company decide? The company desires when to make a application because being granted the automated passenger service permit doesn't mean you have to take out the driver immediately. You can be given a permit on the conditions that you are expecting to take out the driver and that you'll reapply for a permit without the driver's later. I think from a bureaucratic standpoint, most firms do not want to apply for the same permit twice just to take out the driver, but they are very well able to do so. From a government standpoint, they would prefer this application to be reconducted. So there is no baseline, there is no ceiling, and there is no floor, and this will be an active conversation. It's very scale dependent, because if you make a application to provide a fixed route bus service, the scale of evidence and miles of reliability will be a lot smaller than if you want to put in a very heavy vehicle, many of these vehicles on the most busy roads in Britain. So it's case proportionate. Makes sense. Wave and Uber found a way to carry passengers before any driverless permit exists, so Wave and Uber did not use the new APS permit which we talked about earlier, so which route did they take and also why did you take it? Wave and Uber collectively applied for a standard private higher vehicle license, so they are taxi drivers, Uber drivers over at Wave. There's no legal difference between the current service available on the Uber app to take on a supervised autonomous vehicle ride and a standard Uber vehicle ride. For any operator or any person who wants to start getting the license and start driving in London, you need three different licenses. You need a license for the driver themselves. This is different to the driving license, you need a private higher vehicle driving license, so usually this requires a criminal background check. You can be a criminal on drive around but you probably shouldn't be a criminal on drive changes around. There's the operator license which is making sure that there is fine operations for any sort of private higher vehicle service. So for instance, if you want to have a mini cab service, you have to make sure that you have a clear booking page that communications and business processes are friendly to customers, and then you have the vehicle license, which is the TFL assessment of whether the vehicle is suitable for passengers. And here Wave would have had to apply to have their modified Ford Mustangs to get a vehicle license. Each of the 15 cars available today would have their own separate vehicle license. So this would be another advanced MOT, you could describe it as done by TFL instead. So Uber already had the operator license, and they have a very good process to get driver licenses and Wave or Uber would have applied for the vehicle license. Once they got all free, it's legally perfectly fine for those humans to drive or not drive but supervise the driving, but legally they are still liable. Okay, but there would be no different license for the driver because like it's an AI driver more or less that it's just supervised. So it's still the driver in the vehicle that needs to have a license. Yes. And do you think this is more like a shortcut route or is it a natural first step for everyone that comes after? It's an opening. It's a path that is available but doesn't have to be taken and it could make things a little bit quicker and easier. It could not necessarily, it doesn't have to be a shortcut. Any autonomous vehicle passenger service could be opened entirely without considering this legal pathway of private higher vehicle licensing from TFL. After all, if you want to start a service in Manchester, you shouldn't need to have a shortcut for London and TFL to do this. However, evidence of operating in the UK, evidence of operating in London can help in building future safety cases. Hey, between July 2026 and December 2026, there were 15 wave vehicles operating this many times and there were this many incidents. That's very good evidence for a future permit application. That's how it can be helpful. On the other hand, if an operator might not feel mature enough or might make certain mistakes, it could also create negative evidence for consideration as well. Hence why it's not immediately a shortcut, but in most cases, a responsible deployment will help in providing a full driverless license. And they set up as far as I know, Uber owns and operates the cars and wave only supplies the driving software. So why does that split matter for the permits? So in the automated passenger service permit, there will be only one operator, one company providing the application and the evidence and it is their responsibility to manage all third party response and disabilities and relationships. This means that both wave or Uber could make my application and how does it matter that currently Uber has the operating license? Not much necessarily. Wave could very much make that same application. It depends who wants to submit all the evidence required. Uber will have better evidence in some details, especially on the operational side right now, saying, hey, we will have these vehicles driving in this location and we'll have these standards of safety around the depots, for instance, so that there is no malicious intent placed onto those cars. But from a software standpoint, well, we will have way better evidence. Here are our edge cases here is our safety statistics for millions of miles of testing and there isn't a direct relationship because it's a separate legal process. What happens in the private higher vehicle licensing doesn't mean that the same organizations have to make the same applications or use the same evidence even. It's just probably helpful for them to do that. Now let's go to the main event. So what it takes to get a human out of the driver's seat. So we talked now a lot about the automated passenger service permits or the APS permit. Can you explain what it is in plain terms and also maybe why does DBS A issue it and not TFL? So the APS permit is this fast-tracked service for the government providing or structure for any company to provide a automated service for passengers publicly. For profit, for commercial reasons and it may be a private higher-like style service or it may be a bus-like style service. So this is the only legal pathway to do so. It's defined as part of automated vehicles act 2024. It will be the permanent way as to how these things get processed and as mentioned, it is processed by the DVSA. Why is it DVSA and not TFL? Or the DVSA is a national agency, so they'll be able to take in applications that are considered by any place in the UK that apply to any location in the UK. But why aren't local authorities the major sort of consenting and applicant bodies rather than just consenting bodies? Well, in part, it might be that not every local authority has the same bureaucracy, the same capability of assessing this evidence, the national agency will have many experts on transport, many legal experts on hand, the Leeds Council might not have this many experts on hand, so it makes for more institutional sense and more institutional knowledge to be built on the hand of the central government. Also, the central government wants to have control here. They don't want to be [BLANK_AUDIO] advised by a local authority, admitting loads and loads and loads of autonomous vehicles that themselves might have not predicted. So control is very important here. And TFL is maybe slightly frustrated that ultimately they don't have the final control, but they have had some control placed in this structure as well. And every applicant must then answer basically two questions. So can the vehicle drive itself safely and also can the business run as a safe passenger service? So what's in your opinion the harder to prove? I believe it will be the vehicle driving safely is the harder to prove. Often we'll see an announcement that businesses that are already providing safe passenger services are applying. Uber is going to apply most likely bolt or lift may apply in the future, local transport authorities. So stage coach might apply if they ever get a lot of autonomous bus. These are companies with a very long historical track record of legally operating in the UK, meaning that for them that should be the easy part. Maybe now let's switch to the vehicle side. So the VCA assesses whether the vehicle can be listed as self-driving and your article says the car cannot rely on human watching and stepping in immediately. So what does that mean for the remote system? So for a vehicle to be fully defined as self-driving, it has to provide full commercial routes or a very very high standard or maybe entire standard of driving without any direct remote control. Waymo has a structure in which remote operators can provide driving suggestions and the vehicle itself makes a decision as to which route it controls it takes. This is probably what the secondary legislation this permitting scheme had in mind. They do not want the idea of a dry ring wheel being set away from the vehicle and yet controlling the vehicle. So in any circumstance, a vehicle is not fully legally driverless unless it provides evidence that it can drive and it never needs human control to complete a route. How many routes, all routes, does it have to go up Mount Everest completely autonomously? Probably not. That is then probably later set in judgment and it has to be a very high bar. And will companies have to disclose how often a remote staff helped the car so like this in the Waymo model, this is purely on assistance and giving some suggestions but not driving, but do they need to disclose that? So there's nothing explicitly in the UK site which explains that they do have to disclose that. However, from a safety case standpoint, it will almost certainly be expected. So that's my interpretation of the text. In part, because it explains how certain safety cases can be addressed and it might also provide a very important trend line. For instance, if there is a reduction over time in the amount of remote assistance supervision or interventions, then the company will have a more credible case that we can get, they can get to a very low number or to even to zero. Whereas a flat number on increasing number when it rains in the UK or that it hails or the ice gets slippery, well then that will be the sort of evidence that the VCA and the DVSA will consider. Makes sense. And if I research, there's also the operational design domain and there's the area which the permit allows. So maybe just explain for the audience the difference between the two and also who draws each boundary. So I believe the operational design domain will be a concept very familiar to your audience. It is the boundary in which a autonomous driving system can operate and it often used in any application. It's often sometimes even explained to investors as to what is technically capable for a vehicle to do. So it might drive between these two points and it might only be able to drive reliably at this hour or this level of light or this sort of weather condition. The automated passenger service permit area is very similar. It will also have definitions alongside the ODD. So just as the ODD will define location, the service area will define location and just as it defines hours of operation, it will define the hours of operation. By nature, it will have to be a subset. It will have to be smaller than the ODD and therefore the DVSA will be very interested to hear about the ODD from the operators and from the autonomous vehicle technology providers. And that's the difference. The ODD is industry term while the permit area is specifically only a UK term, but will be, you know, a concept used across all licenses, I believe, at least in this level four environment. Yeah. And can a company also expand this permit area mid permit or does it then need a new application when it's one to expand? It will require a new application and in preparation for this, I also researched it seems that will also require new consent from the consenting authority. So not only will they have to resubmit an application with DVSA saying, hey, we want to expand our permit area here and here's the ODD evidence and here's our safety case evidence where this can be possible. They'll also have to do that, not just to the DVSA, but in London's case to TFL as well. So then TFL has to be convinced. By definition, that these two authorities have to be addressed separately, that implies that they'll have separate bars of consideration. One of them might have a judgment that's a little bit harder to pass than another, but maybe they will have the same bar. Well, that is not entirely clear. Then there's also the service side, which is less about the software or some more about the people and we have now the VCA and DVSA and I think they can run in parallel probably and what do you think realistically, how long is it from a full application to a granted permit? Unfortunately, from our full application to a granted permit, it's not just DVSA and the VCA to consider. It is also the local consenting authority. So the VCA probably will move the fastest. They check that the vehicle is built in a safe fashion and that all copies of the vehicle will be built in a safe fashion and they'll assess the software. Once they do the assessment, the VSO will be granted and they might provide some evidence as well for the DVSA to consider. Now the DVSA will also consider the operational side. Ultimately, however, the bar can also still stop at the local consenting authority. In this case, it is TFL. When a permit gets finalized and the permit service area gets finalized, the DVSA will submit that information to TFL. They'll say, "Hey TFL, here's our idea for this new service. There's going to be 15 wave cars and we're going to take all the drivers out and they're going to operate here in North London and this is what they're going to do when there's a power cut and this is what they're going to do when it snows. What do you think about that? Are you happy?" TFL has six weeks to respond. If they don't, they say, "Oh, well, you don't say anything or you're probably happy, Ben." Consent is granted if there's no response. However, if TFL gives a reason, they say, "No." And here's our reason. Have they considered accessibility requirements might come up? Then there'll be a conversation. From what I understand, a rejection with a reason will almost always lead to that application just collapsing, and then a new application would have to be set up, so that application will get rejected. Now, obviously, in bureaucratic terms, that's how it's defined in practice if someone submits one line in rejection. I'm not entirely sure what happens. No, because we don't want to. Is that a good enough reason? I'm not sure. I feel like people will take respect to the system and not play around these edge cases, but the way that the law is written right now, that sort of edge case might appear at some point. However, that is currently the process. How wrong, right? Very politician answer. I still don't give you a time. We don't know. It is the blunt way to put it. There's no sort of base rate. Once we get one application through, they will say, "Hey, we applied eight months ago, and now we got the permit." Then we'll know, "Oh, it probably takes roughly eight months." However, if there was a lot of rejections at once, TFL is unhappy, provides the evidence, it goes from the start. Maybe they have to change the service enough that a whole new operational review has to happen from DVSA again. If they change a little bit, then maybe those first few stages will be repeated quite quickly. But if a lot of changes are required, then that could be delays for another few months. The expectation that a lot of companies set out should be probably delayed by at least two to six months at any stage right now, at least from the current regulatory environment that we seem to be having. And the applications opened in May, but as of now, TFL said, it had received no APS applications or constants requests for London. So why has nobody reached that stage yet in European? At this stage, the process is a little bit opaque. The evidence gets passed from the application. account to DVSA and that is not public information as to how much information has been passed and at what stage before the TFL consenting stage are we at. And TFL did not even have to admit whether they received any applications or not, right? Officially, legally that's also a fairly, you know, not involved with the public process. Why haven't we considered it? I think we would imagine that the operators want to have an application that passes. They want to make sure for a hundred percent that they get the consent and they're happy and they will get through. And in the application process, there are stages which allow for consultation, for conversations with all involved parties, meaning that legally, it's probably fine for these operators to talk with TFL beforehand, at least through certain channels. So they might be having these conversations saying, hey, TFL, this is kind of the idea of what we would want to submit to you. Are you happy with that? And then TFL might say, no, might have already said no, who knows? And maybe that's why we haven't seen a full application formally, because informally there's not yet certainty that they'll be back to go ahead. And any rejection will definitely delay things for, you know, at least a month if not multiple. So it's the operators being cautious and maybe they themselves aren't yet ready to even submit an application. That's another explanation. So once again, there's no set answer here, but there's some hypothesis we could set up. Yeah, it makes sense. And also this six-week look, this probably only works if companies know what TFL wants to see more or less. And TFL has also not published its consent guidance, so to speak. So in July, the director of strategy said TFL was not in a rush. So what is the practical effect on operators if they don't have like this consent guide? Yeah, the consent guidance isn't a legal requirement. I believe this was some industry expectation, you know, quite publicly, maybe some industry players. And here I don't know, but one hypothesis is, hey, TFL, can you like write publicly like what do you want? Because we understand that London's a very complex city and that you want to make sure that public transport that TFL operator is not disrupted by autonomous vehicles. For instance, it's currently unclear whether current taxi lanes or current bus lanes would be allowed for autonomous private higher vehicle services. Some private higher vehicles occasionally in the past get to use bus lanes and these limited roads, but will all autonomous vehicles be allowed that? That should be publicly written because currently it is publicly written for Uber and Bolt and the service operators. The fact that TFL hasn't published this and they've had quite a few, you know, months to consider this over a year, really might suggest that there isn't enthusiasm, especially with that comment as well. So it's an industry observers would comment that maybe TFL doesn't want these services to be happening right now. And they're using all these different mechanisms that are legally formed or that are processed formed, but not statutory, to make this a little bit slower on a process. They ultimately maybe believe that politically they can't stop it this forever, but actually legally, they can actually delay this for a very long time and ultimately for London forever because they are the consenting authority. Politically, however, there would then be other considerations that maybe the DFT would want to place. The Secretary of State has other ways of implementing TFL outside of this formal process. So the Secretary of State really wanted these services to happen. There might be other pathways, obviously, with very, very far from that sort of world. Another interesting point is that the chair of the London Assembly and Transport Committee has said that the centric government has stepped on London's devolved powers. So is there a real tension between the Department for Transport and Transport for London or is the system working as design? Is there real tension or is the system working as designed? You could say both. These are not either or I believe TFL is making clear through the different mechanisms and through the different bodies where it has influence over that the current process might be a little bit too fast or that they don't believe they have the power to define the service as they want. But ultimately, they might not care about having the full consenting and permitting license, right? They might want to just say, hey, earlier on in the process, we want to say that there is some fixed requirements in the permitting scheme. But the demands haven't been made public. So we don't know why TFL is delaying processes or trying to delay processes or being perceived because we're not ultimately saying that they've delayed anything yet, because there's no clear evidence of that. But the industry might point to some tensions ultimately, however, you know, these are individuals. They have the right to save the things in their individual capacity. Maybe there are no tensions at all. And maybe one last question to round this up. So I want to look into the future with you and your crystal ball. So please predict for us the first APS permitting London. So which company roughly when and what will it look like? So how many cars? Maybe what area? So give us your future read. Yeah. So two notes first. First, the Institute for Driverless Transport has no preferences for any commercial success in any of these companies. We are entirely neutral players, but we wish all of them to be successful and beneficial for British society. And secondly, I have written some predictions in the past at self-driving insights. You can find predictions for 2026. Self-driving insights, 2026 predictions is the post. I recommend checking that out. But with that out of way, if I were a personal betting man, what does Tim believe? I believe wemo has a pretty good chance of being the first to have a license. There's no reason for them to not use almost every single car that they already have available. They have a hundred cars testing in London. So I would expect 80, 90 cars or even as many as 200. I think that's the range. I could see 200 cars being in the first license or maybe even 250 or 80. That depends if wemo feels like they want to ship more cars. If there's a very big commercial pressure, if wemo believes that we will get the license two months later or a month later, I would expect more cars. But if there's less just the same amounts of us today, then you could probably imagine that the next APS permit doesn't get granted because we care about market share here. We care about brand, right? So pace of deployment, wemo will set their resources accordingly. After all, they have lots of cities in the United States where they have competition as well. So what else? Where? I publicly made this prediction. It's going to be in North London. Unfortunately, over South London, I say personally, unfortunately, because South London has less transport accessibility in the first place. A lot less tube stations, more buses, but the buses have less capacity overall than tube stations and tube rides. But there's more money to be made in North London. There's more offices, there's more high paying jobs and there's also more demand for existing private higher and vehicle services. So Uber makes more money in North London and North Central London, specifically North of the river, I mean. So I'm expecting places like Kings Cross, I'm expecting places like Paul Bern and St. Pankquist, perhaps even Westminster. It would help in convincing more politicians that this is a very important reliable service because maybe the politicians would start using it, but also I believe the city of London, I think unlikely to see Canary Warf, despite the fact that there's quite a lot of high paying jobs there, partially because the road density isn't great. And also it's quite far from depots. A lot of depots are currently being set in North West London around Park Royal. So you also do not want services way more starting closer to Greenwich and closer to Canary Warf because getting back to the depot would be a long one hour and a half or two hours in traffic drive, which would be quite wasteful in electricity and time. So my prediction 80 to 250 cars, North Central London way more. Thank you very much for that prediction. We will keep you on that. So we will track it and then yeah, revisit your prediction afterwards. Benefits all will get tarnished, but I believe more people should make prediction and we can only become more accurate in our beliefs once we see our predictions fail or succeed. So then we can become more wise for the next prediction. We're at the end. Thank you so much for taking the time. It was very insightful and I highly recommend everyone checking out his sub stack and also his post about the permits and also the post about his predictions. And yeah, Tim, thank you so much again. See you soon. Daniel, thank you. It's been an honor.